Dispersivity - Groundwater Resources Association of California

GRA GRA Proudly Proudly Presents: Presents:
GRACast: GRACast: May May 26, 26, 2011 2011
Dr Dr Dr. Dr. Fred Fred Fred Fred Payne Payne Payne, Payne, ARCADIS ARCADIS ARCADIS ARCADIS
“The “The Continuing Continuing Evolution Evolution
of of of of Remediation Remediation Remediation Remediation Hydrogeology
Hydrogeology
Hydrogeology”
Hydrogeology”
Join JJJoin ii GRA GRA GRA GRA TODAY! TODAY! TODAY! TODAY!
Email questions during the event to
gracast@grac.org
gracast@grac.org
1

Mark Mark your your Calendar: Calendar:
Upcoming Upcoming GRA GRA Events Events
(find ( (find ( out out more more at at www.grac.org)
www.grac.org) gg g) g)
Join JJJoin ii GRA GRA GRA GRA TODAY! TODAY! TODAY! TODAY!
• Groundwater /Surface Water Interaction: California’s Legal and
Scientific Disconnect. June 14, 2011, Sacramento, CA
GRACast: National Groundwater Monitoring Network . June 28,
2011
28 th Biennial Conference and 20 th GRA Annual Meeting. October 5
and 6, Sacramento, CA
Advanced Tools Workshop: November 17, 2011 (more info coming
soon)
Emerging Contaminants: February 2012 (more info coming soon)
2

The Continuing Evolution of
Remediation Hydrogeology
Getting Things Done in a
Heterogeneous, Anisotropic World
California GRA Webcast
California GRA Webcast
May 26, 2011
Fred C. Payne, Ph.D. ARCADIS

Evolving Remediation
Hydrogeology - Outline
Remediation hydrogeology, defined
Hydrogeology Outline Hydraulics and transport
Changing conceptual foundation
Tracing the origins of dispersivity
The advection-diffusion model comes of
age
Mapping transport and storage
Separating characterization from
monitoring
Advantage of real-time, adaptive highres
characterization
Using g improved p characterization to
reduce remedy costs
Saskatchewan River, Alberta
4

Remediation Hydrogeology – An
interdisciplinary practice of environmental
science and engineering that has the
objectives of quantifying and removing or
containing unacceptable contaminants in
groundwater.
Mapping contaminant transport
and storage in aquifers
Quantifying potential for
contaminant exposure
Designing g g control strategies g and
remedial actions that contain or
eliminate the potential risks
associated with contaminants in
groundwater
5

Why is Change Needed?
Difficulties in predicting contaminant migration
Remedies often disappoint
Costs to closure are quite high
Past experience is generating significant
pressure to declare sites as “technically
impracticable”
impracticable
6

If we abandon the simplifying assumptions,
what can we gain?
Improved predictability (although
modeling is a more significant
challenge)
Typically, remedies can be much
more focused, saving money on
remedy design and operation
Remedy effectiveness may
increase significantly
Site characterization requires
increased investment, but:
How do we get the data cost
effectively?
How can we use the data to
reduce remedy costs?
7

The Water Supply Hydraulics Legacy
Focused on very large scale
aquifer productivity
Conservatively estimates
aquifer yield
Based on large-scale averages
and steady-state observations
Works very well for water
supply, however:
Bulk aquifer analysis can
obscure important details
Underestimates transport
velocities
Canadian River Municipal Water Authority
2,250+ gpm

Conceptual Foundation
That allowed linkage of transport
analysis to hydraulics
A series of simplifying
assumptions regarding the
aquifer matrix …
Homogeneous
Isotropic
Gaussian
Steady-State
Well sorted coarse sand, Makaha Beach, Oahu
9

The concept of dispersion was developed as an adjunct to
large-scale groundwater hydraulics, to provide simplified
predictions of contaminant migration
velocity-independent
y p
dispersion (diffusivity)
D = v D*
L L x
velocity-dependent
dispersion
(hydrodynamic
dispersivity)
presumed to be a
minor factor
The Gaussian outcome of the advection-dispersion equation (ADE):
C
0 L vx t vx L L vx t
C erfc erfc exp exp
erfc
erfc
2
2 DL t
DL
2 DL t

Tracing the origins of dispersivity
The old view - “Classic”
transverse dispersivity
Calculated from
mechanical dispersion
coefficients (alpha x,y,z)
that aren’t tied to any site
structure or contaminant
characteristics

Tracing the
origins of
hydrodynamic
dispersivity p y
Danel’s work at
UNESCO (1952) ( ) – he
described a 6° “cone”
in conceptual
dispersion studies…
The visual depictions
associated with that
work k gave a different diff t
impression
28° arc

Tracing the origins of hydrodynamic dispersivity
1972

Freeze and Cherry followed Bear’s lead…
1979

… as did Fetter and others
1980 - 2001

Re-Thinking Dispersivity
Fluid dynamics – random walk prohibited in saturated porous medium laminar
flow systems
Logical (mathematical) analysis – reductio ad absurdum
What do field studies show?
Cape Cod tracer study
Borden aquifer studies
Ft Devens Devens, MA
16

Dispersivity – A Founding Simplification That
Needs Revision
Two sources of dispersion:
Contaminant molecules
randomly spread out in
groundwater and become
more dilute with time – (this is
aqueous-phase diffusion) time
Water was assumed to
randoml spreads in the
saturated aquifer matrix,
further spreading the
contaminant molecules –
(note that water is an
incompressible fluid in
laminar flow condition)
NOT TRUE
distance
TRUE
17

Dispersivity comes undone via reductio ad absurdum
- Extending Pascal’s Triangle to Large Scales
18

Extended Random Walk – Probability of lateral dispersion
becomes immensely improbable
Binn
counts aftter
a 10,000-ball
trial
300
250
200
150
100
50
0
11,000 000 – step lattice
0.5 m travel through 1 mm particles
Average of 100 trials
-400 -200 0 200 400
Outlet channel position, relative to centerline
45 45

DeMoivre-LaPlace Expansion for Extended Lattices
Distance D fromm
Centerlinee
(m)
4
3
2
1
0
0 250 500 750 1000
Distance Travelled (m)
99% 9
99.99% %
99.99999%
If 1 k l th l d lk di i
If 1 kg leaves the source plane, random walk dispersion
drives less than 1 mg beyond 3.5 m from the plume
centerline, in 1,000 m of travel along the plume axis

Field research repeatedly confirms that transverse
dispersivity is near-zero
Borden aquifer studies – Rivett, Feenstra and Cherry
Natural aquifers q show near-zero
transverse dispersivity

Cape Cod Tracer Studies
LeBlanc, et al., 1991
3t 3tracer iinjection j ti wells ll Garabedian and LeBlanc LeBlanc, 1991
656 multi-level
samplers over 200-m
flow path
Garabedian and
LeBlanc LeBlanc, 1991
Transverse horizontal
dispersivity = 1.8 cm
Transverse vertical
dispersivity = 1.5 mm
Hess, et al., 1992
22

Borden Tracer Simulation
Longitudinal dispersion increases with screened interval length – short screened intervals
yield low dispersivity (i.e., longitudinal dispersion is partly an artifact of the screened
interval – the longer the screen, the greater the apparent longitudinal dispersivity)

Implications of Near-Zero
Dispersivity
Contaminant mass transport
is should be controlled by the
aquifer matrix structure. With
very y limited spreading. p g
Contaminants are likely to be
concentrated in a small
portion of the aquifer cross
section
Remedies may be designed
ttottake k advantage d t of f thi this
distribution pattern

Compare:
Dispersive
Prediction
versus
Actual Plume
Development
The
Ft Ft. Devens
Example
Expected plume
migration pattern,
if the standard
dispersivity model
were correct

Actual
Ft Ft. Devens
Plume
Negative
transverse
dispersivity
(plume narrows)
Transport
velocity more
than 10-fold
greater than
average velocity
PCE > 1,000 ug/L
extended more than 3-
fold farther than
projected with
advection-dispersion
IImplications: li ti
- Plumes are long, narrow and more
concentrated
- Dispersion-based search patterns
easily miss the plume

Jan-Feb 2005
Contaminant transport
isn’t random – it’s
controlled by the
hydrostratigraphy.
Ch li i
Channeling is now
commonly reported in
the literature

A New Conceptual Foundation
That Precedes The Conceptual Site Model Development
Homogeneous g
Heterogeneous g
Isotropic
Gaussian
Steady-State
Anisotropic
LogNormal
Perpetual Transient State
28

Fate and Transport in a Heterogeneous
Aquifer – An early publication …
29

Dispersivity Replaced – The Advection-Diffusion
Approach Comes of Age
Transport occurs in mobile
pore space channels
30

Aquifer Matrix
Structure Effects
– Groundwater
and Mass
Contaminant
Mass Transport
Bulk averaging
obscures velocity
impacts of
heterogeneities g
31

Matrix Structure Controls Diffusion Impact
32

Aquifer Matrix Structure Effects – Mass Storage
Two aquifer volumes with equal:
Average hydraulic conductivity
Mobile porosity
Groundwater transport velocity
In the high-mass-transfer
geometry, the rate of f diffusive ff
migration into the low-K zones is
approximately 10-fold greater
than for the low-mass-transfer
low mass transfer
case.
Shorter diffusion path lengths
Increase surface area for
diff diffusive i exchange
h

Major Concept – Diffusion Impacts Mass Transport and
Storage in Real Aquifers
multi-permeability aquifer
diffusive mass storage
TCE storage capacity in a
heterogeneous aquifer
34

Multi-Domain Transport and Storage
35

Multi-Domain Transport and Storage
36

A Closer Look at Heterogeneities
10,000 mg/kg TOC < 500 mg/kg TOC
37

A closer look at flow heterogeneity
Flow (> 99%) Storage (> 99%)
10 -2 cm/sec
1,000 ft/year
10-6 10 cm/sec
0.1 ft/year
10 -4 cm/sec
10 ft/year
38

Key Outcome: Plume Maturation in
a MMulti-Porosity lti P it SSystem t
Development
of mass
storage in
low-mobilityy
pore spaces,
through
aqueous-
phase
diffusion
Early plume
development
Maturing
plume
After source
removal
39

Depth D (z-axxis)
Dissolved-phase plume expansion
High-K, High-C zone
Diffusive mass transfer
from high-K to low-K zones
Low-K, Low-C zone

Depth D (z-axxis)
Dissolved-phase plume contraction
High-K, Low-C zone
Diffusive mass transfer
from low-K to high-K zones
Low-K, Mid-C zone

Heterogeneities
and Diffusive
Storage Have a
Major Impact on
Migration of a
Clean Water Front
Heterogeneous,
anisotropic system
(most natural aquifers)
Homogeneous,
isotropic system
(theoretical and
few laboratory
systems)

Impact of Plume Maturation on Cleanup Times
43

Question & Answer Break
Beaver Island, Northern Lake Michigan
44

A Revised Conceptual
Framework
Heterogeneous
Heterogeneous,
anisotropic matrix
Structured by y depositional p
processes
Becomes tractable
through
Intensified
characterization
Hydrostratigraphic
analysis
45

What We’ve Observed (1) – Groundwater flow (and target
compound mass transport) is often concentrated in a small
portion of the aquifer cross-section
Depthh
(z-axis)
Mass flux distribution is
ttypically i ll concentrated t t d iin a ffew
high-conductivity conduits

What We’ve Observed (2) – Groundwater elevation gradients are
poor predictors of local groundwater flow direction
1 Mile
40°
offset
47

What We’ve Observed (3) – High-definition mapping allows
remedy focusing – significant return on characterization
investment during remedial action
Groundwater flow axis
Invasion front, with
Mature plume area with
mass concentrated in
mass concentrated in
higher-flow zones
lower-flow (especially
interbedded) zones

Three Premises on Getting
Plumes to Closure
Aquifer matrix structure, chemical properties
and exposure duration control how easily we
can get to site closure closure.
For small plumes, intensive remedial action
can often defeat the matrix structure and
exposure duration problems problems, allowing us to
reach a cost-effective site closure.
For large plume footprints, intensive
treatment of the entire footprint is normally
not affordable and we work toward site
closure with a combination of active
treatment zones and zones of clean water
flushing. In these cases, the target
compound, the age of release and the
aquifer matrix structure are all critical
ddeterminants t i t in i how h easily il we get t tto closure.
l

How do we get the data?
Introducing Next Generation Site Characterization – Building Detailed
Physical Site and Contaminant Distribution Mappings
High-Definition mapping in 3-dimensions
Real Real-time time data collection and adjustment of search patterns
High-Definition Mapping Trial
Collaborative Demonstration with
Amsted Industries
50

evisse
design
continnue
operationn
No
Site Characterization
Baseline geochemistry and hydrogeology sampling
No
Remedy selection and initial design
Design support and pilot study sampling
Key design assumptions
validated?
Yes
Final design and full-scale system construction
Full-scale system operation
Key system operating parameters –
technology-specific
SSystem t variables i bl within ithi accepted t d
range?
Yes
No
Site Characterization is
the Key to Closure:
Conceptual Site Model
Development
Remedy Selection and
Design, and
Operational Decision-
Making
modify dif
operation
Troubleshooting based on expanded variables list
Project objectives achieved? Adaptive Adaptive Design, Design, Data Data Discipline Discipline
Yes
and Decisiveness
Remedy complete

Conceptual Site
Model Development
Built incrementally
Data-driven
Matures through
hypothesis yp testingg
Becomes increasingly
quantitative,
independent of
numerical modeling
Continues to evolve
th through h th the entire ti
project life cycle,
including remedy
operation

Remedy Design Support Data
Wh What t we need d … Methods to obtain data …
Target compound distribution Hydrostratigraphic analysis
Mass storage zones
Mass transport zones
Fluid flow characteristics
Natural gradient transport
velocities
Fluid injectability
Aquifer matrix mass transfer
behavior
High-mass-transfer versus
low-mass-transfer
NAPL mobility
Cone penetrometry
Hydraulic profiling tools
Vertical aquifer q pprofiling g
Short, temporary screens
Waterloo profiler
Natural gradient tracer studies
Reagent injection testing
LNAPL tracing

Scale-Dependent
Remedial Strategies
Spectrum
of Sites

Remedial Strategy Drives
Characterization Requirements

Re-Thinking Monitoring Wells – Separate Site
Characterization from Monitoring
10-year life-cycle cost of a single
monitoring well ~ $150,000
(construct (construct, develop develop, monitor and
report quarterly, abandon)
AFCEE study – less than 10% of
analyses l exceeded d d regulatory l t
criteria (all compounds analyzed)
New approach – separate site
characterization from monitoring
well construction – characterize
then determine most effective
monitoring it i well ll llocations. ti
Yields a significant reduction in the
number of monitoring g wells
56

Three-Fold Characterization Strategy
1. Direct mapping framework: geology and target compound
sample collection to build a rudimentary three-dimensional
mapping;
2. Interpolation (extending the mapping between points of direct
observation) and extrapolation (extending the mapping outside
the range of direct observations) – For example,
hydrostratigraphic analysis to form an interpolated mapping from
the explicit mapping points and to guide further sample
collections, ll ti and; d
3. Stimulus-response and hypothesis testing – on-going
refinement of the target compound distribution and hydrogeology
iinterpretations t t ti iin response tto pumping i and d iinjection j ti ttrials, i l
remedy pilot testing and other transient-state system
observations.

Basic Approaches to Building Scale-
Dependent Characterization Datasets
Direct Characterization
Discrete-interval samples p
Continuous lithologic logs
Hydraulic profiling
Interpolation and Extrapolation
Hydrostratigraphic analysis
GGeophysical h i l ttesting ti
Stimulus-Response Testing
Tracer injection
Slug and pumping tests – focus on
transients
Remedy performance

Example: Building and interpreting a high-
resolution site characterization
Source cut off 1975
Look inside the envelope: TCE resides in lower-K zones
Interpretation: trailing edge of long-term washout
High-Definition Mapping Trial
Collaborative Demonstration with
Amsted Industries
59

Mapping at High Definition
Challenge – 10+ yrs of
conventional investigation
Plume mapped using 50+
wells did not resolve site
owner’s owner s concerns concernsabout about
contaminant transport and
past lagoon discharges
High permeability sand
aquifer
Solution – High-Definition plume
mapping to expedite
characterization
Adaptive scope of vertical
profiling to map plume and
evaluate remedy options

Prior Monitoring Well
Network and 3D Plume
Interpretation

High-Definition TCE Profiling

High-Definition TCE Plume

Building a Conceptual Site Model – Original
Source, pre-1975
64

Building a Conceptual Site Model – Clean
water wash-out, post-1975
65

Outcome
Investment
40-acre parcel, 100-ft thick,
mapped at high-res (1,000
discrete samples) for the life
cycle cost of 3 monitoring wells
Return on Investment
Identified unknown plume with
upgradient sources
Demonstrated concentrated
plume: More than 95% of the
mass discharge occurs in less
than 10% of the cross-section
High-definition plume mapping
will focus remedy

Putting the Evolved Hydrogeology to the Test ….
33-mile-long mile long TCE plume plume in
the Ogallala Aquifer (alluvial
fan, 100 to 150 feet bgs)
Performance based
contract – MCLs in 10 years
Principle p challenges: g
Find and intercept TCE in
transport
PPotential t ti l TCE storage t in i
the dispersed plume
Former Reese AFB, Texas
67

Former Reese AFB Environmental Sites
Solvent Still Area
Southwest Landfill
Tower Area Plume
Petroleum, Oil and Lubricants (POL) Area
Plume Status - 2004
68

2005

2007

ERD Treatment Sector
2008

2009

2010

2010

2011
75

Why are MCLs Possible at This Site?
Site Hydrogeology
No DNAPL was found below the aquifer surface
Vadose zone source mass was completely removed
Fast flowing, low-mass-transfer groundwater system
Diffusive interaction limited – no significant mass storage effects outside
the source zone
Site Characterization
Hydrogeology and conceptual site model continually re-assessed
Remedy adjusted as-needed
Regulatory
Allows flexibility in site operation
Collaborative environment
Relations with Adjacent Landowners
Gaining site access where it was previously denied
Maintaining good working relationships was critical to the success of this
project p j
76

Recap
Remediation hydrogeology is transitioning away from its older
simplifying assumptions that were associated with hydraulic
anal analysis. sis
Recognizing the heterogeneous anisotropic nature of aquifers, it
clear that site characterization efforts need to be intensified.
Separation of monitoring and characterization processes is critical
to cost-effective characterization and effective monitoring.
IInvestment t t in i site it characterization h t i ti can yield i ld several-fold l f ld returns t
through optimization of monitoring and cost reductions for remedy
installation and operation.
Real-time, high-resolution site characterization methods are
developing rapidly.
77

Beaver Island, Northern Lake Michiga
78

Join Join GRA GRA TODAY! TODAY!
Send your questions to this presentation
anytime to gracast@grac.org
For more information on GRA
including membership
opportunities, pp , pplease
see our
web site at
www.grac.org
Please help us make future events better by providing us feedback
at our online survey at http://www.grac.org/gracastsurvey Last page